Skip to main content

Table 3 Calibration of the existing time-dependent model with and without the trend indicators

From: Does adding risk-trends to survival models improve in-hospital mortality predictions? A cohort study

  

Existing model with trend indicators

Existing model without trend indicators

Risk Decile

# observed deaths

# expected deaths

z-score

p

# expected deaths

z-score

p

1

5

4.31

0.3298

0.7415

5.37

0.1609

0.8722

2

9

12.75

1.0498

0.2938

13.68

1.2664

0.2054

3

21

24.74

0.7527

0.4517

26.03

0.9853

0.3245

4

36

43.45

1.1298

0.2586

44.59

1.2861

0.1984

5

54

68.90

1.7952

0.0726

69.50

1.8597

0.0629

6

103

108.72

0.5482

0.5836

109.97

0.6646

0.5063

7

174

171.65

0.1796

0.8575

173.49

0.0386

0.9692

8

260

285.35

1.5008

0.1334

285.02

1.4821

0.1383

9

453

474.99

1.0090

0.3130

478.93

1.1850

0.2360

10

1525

1496.72

0.7309

0.4649

1497.23

0.7177

0.4729

Total

2640

2691.59

0.9943

0.3201

2703.82

1.2274

0.2197

  1. We divided the validation admissions into risk deciles on each admission day (based on the patient's risk score for that day from the model with the trend indicators), determined the number of observed and expected deaths from each model within each risk decile on each day (where the daily number of expected deaths was equal to the sum of the daily hazard of all patients within each decile), and finally summed the number of observed and expected deaths within each risk decile across all admission days (shown above). We tested for a significant difference between the observed and expected number of deaths within each decile by calculating the p-value associated with standardized z-statistic, where z = (observed-expected)/(√expected).